Rethinking Enterprise Knowledge: Discovering the Data Bounty of Documents
Knowledge for the modern enterprise is not as cut and dry as one might think. In fact, knowledge is often defined by systems, not by its use or users. Yet in this age of artificial intelligence, knowledge is more than power; knowledge is the continuously renewing source of context and intelligence that AI outcomes demand.
This is where documents—the spreadsheets, Word and Powerpoint files, brochures, and PDFs—start to play a new role, especially for customer experience (CX) leaders wondering just how they will continue to power generative AI models that are hungry for the content and context that knowledge delivers.
Documents are a funny conundrum for modern companies. From brochures and presentations to reports and technical documentation, from spreadsheets and summaries to agreements and quotations, documents can be the catalyst to creation and the origination of obsolescence. They can also be easily overlooked once the final version is approved and used…relegated to cold storage and forgotten. All the work, the collaboration, the insights, the wisdom, and the institutional knowledge can be lost.
The age of AI has issued a call to action for documents across every organization. In a time when AI agents search for more data to satiate their hunger, documents become micro-repositories of knowledge and context, but only when we stop thinking of them as a final destination.
AI has business leaders rethinking fresh, renewable data to represent business, operational, employee, and customer context. Much of the agentic conversation has focused on the capacity of AI to generate content, leveraging existing documents and insights about markets and customers to create new assets to suit an audience of one. This is just the first step from the starting point. To truly capitalize on the AI + documents opportunity, it's time to think differently.
Here are three steps to rethinking what comes next:
Step one: Think about documents differently.
Smart organizations have already established strategies that collect, secure, and store explicit knowledge—the final documents and documentation of codified information. Teams leading with AI have also thought about documents as the output from agentic workflows, empowering content tools to generate new assets autonomously. But to shift from documents as passive assets to active AI fuel, we need to think beyond what happens to an asset after the fact and shift the continuum to a more continuous cycle that taps into the tacit knowledge (think of this as the knowledge of HOW the sausage gets made) that contributed to decisions and wins.
The work that went into documents should power new strategies, new answers, and new opportunities. Thinking about documents as a form of renewable energy opens opportunities to think about continuous improvement, not just of that document but of the workflows attached.
Take, for example, how documents impact the sales process. From the customer's initial quote to the end statement of work, documents carry the experience end to end. But the real knowledge is hidden in the requests, comments, edits, strikethroughs and additions. If we think about these documents as active assets, knowledge about business process, workflow and the customer can be extracted and set to work, helping sellers better enter renewal conversations armed with knowledge and ready to materially shift the experience. Looking beyond the final and thinking about documents as a lifecycle allows systems to operate on context and intelligence, not passive reaction.
Step two: Think about knowledge differently.
Much like shifting away from documents being a passive asset, organizations should think differently about knowledge. As an example, within the context of customer service, a knowledge base is often established to help service agents resolve key issues. This is less about documenting knowledge and more about embedding knowledge that directly impacts performance and experience.
The power in embedded knowledge is that it collects the proprietary processes and operational workflows of businesses. Anything less degrades trust. To trust in any AI output, there need to be guardrails, safeguards, and continuous reinforced learning and confirmation. One of the best ways to proactively assess and understand if the output captured in a document is accurate is to have AI trained, tuned, and reinforced by a fresh, renewable source of intelligence.
Let's revisit the example of a sales statement of work. No two sales contracts are the same. They shift and update based on the requirements of the customer and the business. Knowledge fueled by document intelligence can surface recommendations for updates while reducing the friction of manual document reviews.
Step three: Put documents to work differently.
Access to enterprise documents, just like data warehouses and enterprise knowledge bases, tends to be far more limited than expansive. Who can access documents is limited, and rightly so. The idea of cross-organization collaboration, creation, and distribution of documents is safeguarded. But sometimes, the natural ebb and flow of functional work can create artificial limitations.
Once again, let's revisit that sales contract: Does marketing have access to those critical quotes, comments and contract outputs? Does service? Who should be able to ask five years’ worth of quotes and contracts where and how variants have impacted end results or revenue? Is it a sales-only question? Legal? Finance?
Traditional organizational chart-based thinking would say that these functions don't need access to the potentially sensitive contracts or quotes now stored as institutional documents. Access and authorization are necessary, but don't erase the truth that each one of those contracts could unlock intelligence critical to a marketing or service decision when engaging with that same customer.
The suggestion here is not to do away with the security, safeguards, and controls that any responsible document or asset management strategy would include. Instead, this is more about redefining where, when, and how intelligence in a document can be accessed and turned into action.
Revisiting that sales contract for the last time, in a rethought document lifecycle, a signature is not the end, but rather a call for that document to go back to work. Marketing receives signals based on insight from the client's comments that the key message of the last webinar did not land as expected. Service systems are enriched with knowledge that binds CRM, customer record, sentiment, and original need state articulation directly from the customer.
These three steps aren't the only steps…they are the starting point. As an individual or organization starts to ask these new questions, it will invariably invite new questions around the platforms and solutions that can bring this to life. And this is where a real word of caution comes to play: New systems might NOT be the best systems to deploy in these knowledge investigations. Sometimes the answer already exists on a user's desktop or within the enterprise stack.
New capabilities from Adobe Acrobat Studio bring the power of its newly launched Knowledge Base directly in Salesforce Slack or Microsoft Teams, bringing knowledge directly into organizations' work.
This rethought world of documents and the new conversations true enterprise-wide knowledge can amplify and accelerate can drive real business velocity and unlock the output all leaders seek in exponential growth. The knowledge and intelligence from documents will power the continuum of data to decisions thanks to data that is derived from decisions. So, here's to putting documents back to work, not for the sake of more work, but rather for the sake of more successful outcomes.
Liz Miller is a vice president and principal analyst at Constellation Research.